Weather or Not
As March 8 approached, local forecasters in Denver had a feeling the day wouldn't be right for sunbathing. The temperatures seemed likely to be on the brisk side, and there was even a chance of scattered snow showers. But neither the prognosticators at the National Weather Service nor the weather professionals at the city's local TV stations -- the folks whose predictions largely determine what clothes people put on before heading out the door each morning -- expected any big deal.
Wrong. Early that afternoon, a wind and precipitation blitzkrieg struck the metro area, producing what Channel 9 meteorologist Mike Nelson calls "two hours of hell on the highways -- a snow squall that was almost like a summertime thunderstorm."
"It looked like a blizzard for a couple of hours," says Bob Goosmann, the chief predictor for Channel 31. "I was at the mall with my wife, and I just kept mumbling, 'I didn't think we'd see this much snow...' I thought we'd drop into the thirties, and we dropped into the low teens. I was dumbfounded by the whole thing."
Marty Coniglio, the A-list weatherman at Channel 7, was also caught off guard. "I worked that whole damn day, and at one point, I went into the weather office and thought, 'I got hosed. I just got screwed by two computer models.'"
So did Denver drivers. The Colorado Department of Transportation, which relies on much the same weather information as the citizenry at large, was totally unprepared for a major weather event; for that reason, Nelson notes, "they didn't have time to put down any mag chloride," a chemical that helps prevent the pavement from freezing. Playing catch-up proved difficult: Kieran Nicholson, writing for the March 9 Denver Post, reported that "about 50 CDOT plows were sent across the metro area as the storm rolled in, but they soon got caught up in typically heavy Friday afternoon traffic."
The traffic wasn't typical for long. The highway system was paralyzed for the entire Friday rush hour, doubling or tripling commute times in most locations and turning the T-Rex construction zone into a nightmare on ice. But by far the worst place to be was Interstate 76 near East 74th Avenue, where a pileup of around forty vehicles closed the route in both directions and led to numerous injuries and one death.
"That's the tough part of the business," Goosmann allows. "We'll predict a little bit of snow and say driving could be tricky, but then all of a sudden, you'll get a big storm and a huge pileup, and someone dies. You can't blame yourself for something like that, but it's always in the back of your head. You wonder, 'Was that person watching me last night?'"
Quite possibly. Late-night news roundups garner more regular viewers than just about any other locally produced TV program, and internal surveys and focus groups staged by outlets regularly confirm that a sizable percentage of their audience, and often a majority, watch primarily to find out what the next day will bring.
But how often are forecasters right? Back in 1992, Westword tried to find out by tracking the performance of Denver's top talents over a period of more than two weeks -- and the final count showed nearly as many misses as hits. What's more, the scores weren't that much higher when the procedure was repeated ten years later (see Weather by Numbers).
Nonetheless, Denver's TV-weather gurus -- Nelson, Coniglio, Goosmann, Channel 4's Larry Green and Channel 2's Dave Fraser -- come across as sincere types who seem genuinely interested in getting things right. Furthermore, they believe they do so more often than not.
"The forecast 48 hours and closer is much more accurate than it's perceived," Fraser says. "I'm constantly surprised by how good it is. Even after thirteen years of doing this, it's still amazing to me that we can tell you that it'll be 71 degrees two days from now and be right on the nose."
At other times, they're right on the ear, or maybe the eyebrow: Take April 19, when a sloppy snowstorm that everyone saw coming the night before never materialized. But all five insist upon taking responsibility for botched forecasts.
"I've never met anyone who does this for a living who didn't believe deeply in what they're doing and didn't try to do their best," Coniglio says. "I've found that the most wicked interaction you get with people is via e-mail. Because it's anonymous, the things that are written to you and about you are sometimes quite disturbing. It really is eye-opening. But even then, the things they say after you've had a poor performance, we've already thought about ourselves."
Of course, TV predictors aren't always at fault when they fall. They're simply victims of a system that's noteworthy but far from perfect.
According to Green, the tools of forecasting have improved greatly over the past decade -- among them better and more useful Doppler radar, which he says works best "when you've got a storm right on you. It shows right where everything is and lets you put out more accurate bulletins, watches and warnings."
More important from a long-range perspective are assorted computer models offered by a variety of vendors, including the U.S. Air Force and the U.S. Navy. But having extra options isn't always a blessing. "There may be four to six different models that come down in a day," says Green, "and you have to figure out the biases of each model. For example, one model might be biased toward bringing precipitation into the area, and another one might be biased to pressure changes that would affect temperature."
These biases are present even though all of the computer models are using pretty much the same information. "The National Weather Service provides data to any forecaster in the country," Green points out. "I wouldn't want to say we're stuck with them, because they're a very valuable outfit. But they're the only game in town."
To complicate matters further, each computer model decodes the data differently, putting forecasters in the position of interpreting the interpretations. "We have to take these little pictures and put it together to make a mosaic," Fraser says. "And it's up to us to not only learn, but understand these computer biases. I'll bet if you put the five of us in a room, each of us would come out with a slightly different forecast. There would be subtle differences, because each of us will have reached a conclusion because, in certain situations, we trust one model more than another."
There's also the matter of the Denver area's size and scope. The city's official temperature readings are recorded at Denver International Airport, which is practically in another state and frequently experiences vastly different conditions than more populous parts of the city. During a late March broadcast, Fraser noted that the temperature at DIA was a stunning 19 degrees colder than the reading at Channel 2's Tech Center facility. And because of variations in altitude and level of development (concrete tends to hold heat), the temperature in, say, Highlands Ranch will almost always be considerably different from the temp in Evergreen. Stations deal with this discrepancy through the use of maps showing projected temperatures in many suburban locales, but most eventually summarize their forecasts using just one figure. "You don't want to have too many numbers on the screen," Nelson says.
This may come as news to Denver viewers. After all, local stations appear to love flashy imagery that may make little sense to the average layman but suggests a level of expertise among forecasters that reinforces their reputation as all-seeing, all-knowing weather experts.
Consider the March 19 weather segment at Channel 9. It began with a wide shot of Nelson and anchors Adele Arakawa and Jim Benemann before cutting to a "mystery time lapse" -- flashing lights against a dark backdrop that turned out to be illuminated slope groomers at Vail ski area. Next followed a map of Colorado labeled "Highs" onto which numbers flew in from all directions. A video dissolve to a state map labeled "Satellite" replaced the temperature readings with approximations of clouds and fronts; a second dissolve to the "FutureCast" included flocks of animated arrows intended to simulate wind direction. An instant later, Nelson appeared in the "9 Back Yard" only to be covered by a time-lapse shot of Denver that was overlaid with graphics labeled "Statistics" and "Currently at DIA." After a cut, Nelson was seen in front of a series of maps of the continental U.S. and Colorado on which digits and representations of weather systems whipped back and forth like fireflies. Finally, one more batch of graphics -- "Tonight," "Tomorrow," "Planning" -- preceded a goodbye from Nelson, wrapping up a presentation that took just over three minutes. Whew.
Coniglio gets about the same amount of time as Nelson does for his weather segments, and he says it's not enough to go into many details. "You begin and end with a little chitchat, which people say they hate, but it's just a natural segue from one thing to another. So you lose thirty seconds there, and you lose another thirty seconds doing 'Today' and 'Tonight' -- the stuff you absolutely have to do. That leaves a minute and 45 or two minutes to talk about a state that covers 104,000 square miles -- and to talk about the weather over a period of seven days."
Channel 7 is the only Denver station that offers forecasts seven days in advance, and Coniglio readily grants that its spot on the dial has everything to do with the reason why. But since the reliability of forecasts goes down with each passing day, particularly during volatile seasons like winter and spring, this feature regularly tests Coniglio's mettle.
"There are times of the year, like in June or July, when I'm fine with the seven-day forecast," he says. "But in January, when your day-seven weather is on the Korean peninsula, you're covering an awful lot of geography. You literally have to morph weather events long before they even make it to the mainland -- and if you have a frontal boundary that's a hundred miles farther north than it seemed it would be a week earlier, that's the difference between a high of 55 and a high of 15, which looks like a colossal failure. In some ways, it's like doing brain surgery with a steak knife and a serving fork."
Fraser, who offered seven-day forecasts at previous stations before his arrival at Channel 2 last year, concedes that gazing a week into the future isn't easy, but he thinks it can be done effectively: "Would I plan a balloon launch where I needed so much clear sky and a calm wind based on a seven-day forecast? Probably not. But if I was planning a family picnic and four or five of those days ahead of it looked good, I probably would." Nelson, on the other hand, thinks even five-day forecasts are "pushing it" and is happy he can stop there -- especially since, several years ago, he nearly wound up inching out even further onto the prediction precipice.
"An earlier news director we had, Dave Lougee, came back from a trip to Houston, and he said, 'They do a nine-day outlook on Channel 9 in Houston,'" Nelson remembers. "I said to him, 'Dave, what were the forecasts?' And he said, 'Well, every day, they were the same.' And I said, 'They're not all the same here.'"
Indeed, Nelson and Green, the longtimers of the Denver weather group, talk at length about how difficult it is to predict weather in Denver because of the Continental Divide, which blocks or diminishes some storms and causes others to bounce off in unexpected directions. Goosmann's not sure about that: He says he found it tougher to forecast at previous stops in Providence, Rhode Island, and Richmond, Virginia, because of the effects of the Atlantic Ocean, and in Dallas, which got oodles of storms that had little to slow them down. But Coniglio thinks Goosmann may change his tune. "From his perspective, that's true, but he's basing his observations on having been here for two of the driest, warmest winters in the last twenty years. It's like going up to bat one time and you hit a home run, so you're batting a thousand."
The forecasters have different theories about how to handle broadcasts after they've struck out. Whereas Coniglio tends to make brief mentions before moving on, Goosmann, in particular, goes out of his way to be up front when his guesses bear little resemblance to what actually happened.
"I think you owe it to the viewer," he maintains. "I just try to explain what happened so that they can go, 'I understand -- the front was three hours late.' Or sometimes, when I don't really have an excuse, I'll say, 'We just flat-out blew it.' Because nobody's perfect."
Such honesty flies in the face of the way most newscasts market their weather features. Channel 2 boasts about its exclusive "Pinpoint" forecasts -- a reference to a specific weather device, Fraser says, but also a term that implies precision that's achievable only intermittently, given the current state of forecasting.
Likewise, Channel 31 calls its predictions "On Target Weather," a moniker that was initially underlined by a nightly graphic showing a target with the previous day's high atop a bull's-eye and the station's predicted high either in the same place or, more often, among the rings fanning out from the center, depending upon how close it came. After a few months, the station ditched this attention-grabber, to Goosmann's relief.
"I begged and begged to get rid of that," he says. "A lot of people liked it, but as far as I was concerned, people would remember more when you were off than when you were on. If you were off by one degree or three, they wouldn't notice, but on that occasional ten degrees, they'd be like, 'I remember that.' And that's not what I want to stick in people's heads."
In the end, the "On Target" gimmick did precisely the opposite of its intended purpose, reminding people of forecasters' fallibility instead of portraying them as invincible. That's why stations in general prefer graphic presentations and promotions that avoid acknowledging the gap between possibility and certitude. Outlets are aided in this respect by fancy appliances such as new-generation radar that lets weathercasters depict storm cells at street level in real time. But these gadgets are more adept at showing what's happening at present rather than foretelling what might take place down the line -- and viewers who fail to grasp this distinction are frequently left with unrealistic expectations.
"When people say, 'Is it going to rain tomorrow?' that's not what they mean," Coniglio says. "They mean, 'What time is it going to start raining, when is it going to stop, what time will it rain in Littleton and Thornton, and will I have wet streets when I drive to Lakewood?' Now, we can tell them all that a lot more often than we used to, but we can't always -- and they don't always understand why not."
In retrospect, Coniglio has a pretty good idea why he and his peers booted the March 8 forecast. "It was a really tightly packed front, and the front slipped down further than we expected. It was only about thirty miles off from where we thought it would be; at the same time it was 12 degrees in Cheyenne, it was 50 degrees at Monument Hill [between Denver and Colorado Springs]. But that thirty miles makes the difference between a couple of snowflakes and terrible fog and all the roads icing over."
Adds Nelson, "People often have the mistaken idea that weather is an inexact science, but that's not true. Weather follows all the laws of physics. But it's so complex that we have a difficult time forecasting how a storm is going to move -- and everything can change if there's even a fifty-mile variance on a storm that tracks 3,000 miles across the country. So storms like the one on March 8 happen fairly often; they just don't happen during Friday afternoons in Denver.
"Remember the big Christmas blizzard we had in 1982? Well, there was a storm equally as bad in 1983, but it hit Watkins and Deer Trail out on I-70, so nobody remembers it. Timing is everything."
Get the Weekly Newsletter
Our weekly feature stories, movie reviews, calendar picks and more - minus the newsprint and sent directly to your inbox.